Abstract – In a large scale wireless sensor networks, monitoring
particular area requires large number of nodes to be active. As
the numbers of nodes are more, the energy consumed by each
node will be more. Energy is the most critical issue in wireless
sensor networks. At MAC level, duty cycle scheduling is being
used to improve energy efficiency and a large research is being
still going on how to schedule the duty cycle. The duty-cycle is a
process that frequently repeats sleep and wake periods. In this
paper, to minimize energy consumption the sleep time of node
will be adjust in such a way that it should get active only when
there is an event in the network. The aspiration of this research is
to design adaptive duty cycle algorithm for wireless sensor
networks that minimizes energy usage. The algorithm will work
over both scenario i.e., high as well as low traffic. Due to its
adaptive nature, the energy consumption will be lesser which will
help improves life span of network.
Index Terms’ Duty cycle, Wireless Sensor Networks, Energy
consumption, MAC
I. INTRODUCTION
Wireless sensor networks (WSNs) are generally consists
of a large number of battery-powered sensor nodes that are
deployed randomly. Therefore, at the MAC layer duty cycle
method is used to properly manage energy consumption.
Minimizing energy conservation in a wireless sensor network
is the most essential issue at the MAC level, where the dutycycle MAC protocol is used in mainly cases.
Duty Cycle is a method which reduces the number of idle
nodes in a WSN to achieve energy efficiency. In duty-cycling
networks, sensor nodes are not always active. To forward one
data packet, the sender needs to hold the packet and wait until
the receiver becomes active, which leads to a relatively long
per-hop transmission delay. Such a unique per-hop delay in
duty-cycling WSNs is known as sleep latency. Due to this
latency, sensor nodes may not get adequate bandwidth to
transmit packets in time, and a data packet may suffer an
extremely long delay during the delivery from the source node
to the destination node. Second, due to duty cycling property,
multiple sensor nodes may at the same time transmit packets
to the same receiver during its brief active time, which may
incur heavy packet collisions leading to further increased per hop latency. The problem can become even worse if 1)
multiple data forwarding tasks with different sourcedestination pairs coexist in the network and each of those tasks
has its own time constraint, and 2) interference hinders
concurrent transmissions of nearby wireless links.
A. Sources of Energy Inefficiency
Packet collision: It can occur when nodes do not listen to the
medium before transmitting. Due to this the packets gets
collided, become corrupted and needs to be retransmitted. This
causes unnecessary energy waste.
Control packet overhead: Control packets are necessary for
successful data transmission. They do not, however, represent
useful data. They are very short.
Idle listening: Idle listening occurs when a node listens to an
idle channel waiting to receive data.
Over emitting: The node sends data when the recipient node is
not ready to accept incoming transmission.
This paper is structured as follows. Section I introduces
the reasons for energy wastage in wireless sensor network.
Section II presents the related work on adaptive duty cycling.
Section III presents the problem definition in existing ones.
Section IV presents the design of our proposed ADC
algorithm and analyzes the performance of the ADC
algorithm. Section V presents the implementation part
completed so far. And this paper is concluded in section VI.
II. RELATED WORK
A.ADC Feedback Controller
In [1] Heejung Byun, Junglok Yu proposed an adaptive
duty cycle feedback controller in which duty cycle for wireless
sensor networks is completely adaptive. They designed a
queue and observe the performance under variable traffic
changes. They compared their algorithm with UMAC and
DutyCon. When there is low traffic in the network, the queue
does not get full faster. So, the problem of higher delay is
observed in which UMAC performs superior than the
proposed methodology. But in other cases except this, it
improves both energy efficiency and delay performance by
adapting the duty cycle properly under network changes. But,
International Conference on Convergence of Technology – 2014 978-1-4799-3759-2/14/$31.00??2014 IEEE 1
every time there is a restriction on parameters like number of
hops, number of packets in which UMAC performs better than
the proposed algorithm.
B.DutyCon
In [3] X. Wang, G. Xing, and Y. Yao proposed a duty cycle
controller which improves the end-to-end communication
delay while achieving the energy efficiency. DutyCon decays
the end-to-end delay constraint problem into a set of singlehop delay constraint problems. The each node’s duty cycle is
determined based on the actual packet delay, calculated using
time stamps. DutyCon performs better for single-hop
networks. But it does not have effective control on the end-toend delay under alteration in network condition.
C. Adaptive Duty Cycling (ADC)
In [4] Myungsu Cha, Mihui Kim proposed the ADC
approach that ensures balanced energy consumption of nodes.
The basic idea of the proposed scheme is to group the node
using the topology control and then apply adaptive duty-cycle,
depending on the group size. Simulation using GAF and CPA
as the basic topology control protocols, shows applying the
proposed scheme to these schemes improves network lifetime
at least 25%. It does not enable the flexible approach of ADC
in accordance with the amount of traffics in the network.
D.PMAC
In [2] author designed a new PMAC (Pattern MAC)
protocol, in which sensor nodes itself dynamically determines
the active and sleep time of the nodes. The schedules are
determined based on the traffic of both the node itself and its
neighboring nodes. They compared PMAC with SMAC in
which PMAC outperforms than SMAC in terms of energy
efficiency.
While PMAC was developed to settle in to traffic changes by
raising the number of sleep slots adaptively, it does not lessen
idle listening inside the active slots. Besides, when there is
high contention in the network, using fixed time slot may
create much collision.
E. UMAC
In [5] author proposed U-MAC (Utilization-based MAC)
in which a utilization function is used by each node to tune its
duty cycle. It does not adapt the same duty cycle for each
node. In this, the utilization function refers to the ratio of the
actual transmissions and receptions performed by the node
over the whole active period. The utilization function must be
less than 1. When the utilization function is low then the node
is experiencing a long idle period within the active period. In
Utilization MAC a node maintains two values, Umax and
Umin, and adapts its activity period to let its current utilization
function in this interval. UMAC share the same drawback as
that of TMAC.
F.EAALPL
In [6] author proposed a method in which a node itself
decides to come into active mode according to its local state
because they are not aware of the states of its neighboring
nodes. Initially each sensor sends periodic route update
messages to announce its presence and state. The graph of
active nodes is formed and they direct towards the base station
by determining its descendents coming in the path. Based on
the descendent nodes, the duty cycle is determined. When the
number of child nodes becomes higher, the burden is shifted to
another parent node so as to decrease the energy consumption.
G. Adaptive Sleep Discipline for Energy Conservation and
Robustness in Dense Sensor Networks
In [7] author proposed an adaptive sleep schedule to
provide robustness to the variations in network connectivity.
Each and every node by observing local surroundings decides
their own sleep and active time. The algorithm does not
maintain any information of the neighboring node’s state.
Previous work [10], [11] and [12] could also be useful for
getting various ideas regarding duty cycled wireless sensor
networks.
III. PROBLEM DEFINITION
Energy is the most fundamental issue in wireless sensor
network. In existing protocols like SMAC a fixed duty cycle is
used in which node have to be active during its awake period
even though there is no event in the network. So unnecessarily
the energy gets wasted. So there is a need to adapt duty cycle
in such a way that node should get active only when there is
an event in the network. The TMAC improves SMAC but
again there is problem of early sleep. i.e., the node goes to
sleep mode even if its neighboring node have something to
send to it. Most of the work has been done on adaptive duty
cycle and still going on. In [1], duty cycle is controlled by
queue management. In this, it has been observed that if traffic
load is light, the delay is more because it has to wait for queue
length to reach at threshold value. The comparison is done
with UMAC and DutyCon. DutyCon does not have effective
control under high traffic load. So, in this paper the algorithm
is designed for both scenarios i.e., low and high traffic and
will try to improve network lifetime and energy efficiency.
IV. PROPOSED METHODOLOGY
The complete flow of the proposed methodology is
shown is fig1. The threshold value for energy as well as
traffic is set initially and accordingly the operation has to be
performed as discussed in fig1. If packets are greater than
20 then it is considered to be high traffic or otherwise it is
low.
International Conference on Convergence of Technology – 2014 978-1-4799-3759-2/14/$31.00??2014 IEEE 2
Start
Create network
Assign threshold value for
energy & variable traffic
Go to 2
Go to 3
Apply ADC algorithm
Report data to BS
Results on energy consumed
against no. of nodes alive
under variable traffic
Stop
2
Node energy >
threshold
energy?
Be the cluster head &
broadcast data
Send request to
join a cluster
3
If pkt > 20
or < 20?
High traffic
Low
traffic
CH broadcast
‘hello’
Node buffer
contain pkt?
Be in sleep
state
Come to wake
state
Reply to CH & start
communicating
Buffer free
or have pkt?
If CH changes after
sending hello
Member nodes communicate with
recent CH and discard old CH id
Elect next highest
energy node as next CH
Each node having pkt in
its buffer send ‘hello’
Use AODV for
communication among
nodes having pkt
Node buffer
contain pkt?
Be in sleep
state
Come to wake
state
Come to wake state
after every 10ms
4
4
Energy
Traffic
No
Yes
< 20
> 20
No
Yes
Free
Pkt
e
No
Yes
Continue from 2
Fig1 Data flow diagram of proposed methodology
International Conference on Convergence of Technology – 2014 978-1-4799-3759-2/14/$31.00??2014 IEEE 3
Awaked node
Sleep node
Cluster head
Base station
Fig2. WSN consisting of wake and sleep nodes
The scenario of execution is shown in fig2. It consists of
routed path through active nodes. The simulation parameter
used in this paper is to improve energy efficiency and network
lifetime under variable traffic rate. The proposed workflow is
shown in fig1. First the network module is created, and then
clustering approach is used. Initially every node will have 100
joule energy. When simulation is started, the nodes having
highest energy amongst all will be elected as the cluster head
and the remaining node acts as cluster member. Because of
clustering the workload on every node will not be there, the
burden will be given to one of nodes that will be cluster head.
So, energy of nodes other than cluster will be saved.
Whenever there is traffic in the network, the AODV protocol
will be used for routing.
AODV [8] is an adhoc on demand routing protocol in
which routes are created only when they are needed. The
information regarding sequence number and next hop to the
destination is maintained by routing table. The updated
information about path as well as active neighbors is notified
to all other nodes in the network. The communication will
takes place from one cluster head to other cluster head and so
on. So, multihop communication is supported and energy
consumption will be less after adding duty cycle to it.
In duty cycling, two states are there, sleep and active [9].
In sleep state, a sensor node does not have any task to
complete and energy consumption is much less. In active state,
the nodes participate in data routing and actively take part in
receiving and transmitting as well. The interval of 0.05ms is
set by timer in which it will be active for 0.001ms only and
remaining time it will again goes to sleep state.
Certain assumptions made are as follows:
‘ Initial energy of each node is100J
‘ If packet is less than or equals to 20 it is assumed to
be low traffic and if it is greater than 20 it is
considered as high traffic
‘ CH is said to be cluster head
‘ CM is said to be cluster member
The following code will be executed on the cluster head only.
It will be adaptive because clustering is dynamic. The control
message will contain the source id, reception id, current wake
time and next wake time to get active at the same time.
Pseudocode for low traffic algorithm
1.??CH(i) ‘ N send ‘hello’ to CM
2.If buffer(CM) have pkt
then send data ‘ CH
end
3. if buffer(CM) empty
then go to sleep state
end
4.if CH send hello & energy drain
then CM join CHnew
5. CH1 sends data ‘ CH2””CHn
6.CHn ‘BS
The following code will be executed on the cluster head only.
It will be adaptive because clustering is dynamic. The control
message will contain the source id, reception id, current wake
time and next wake time to get active at the same time.
Pseudo code for high traffic algorithm
1.??n ‘ N if buffer(n)’pkt
then send ‘hello’ to other nodes
2.Use AODV & start communication
3.buffer(empty) ‘sleep state
4. timer on every 10ms go to 1 and continue
5. If energy (node) ‘ 10J
then sleep forever to avoid disconnection.
end
V. IMPLEMENTATION
The simulation is carried out in the area of 1000 x
1000m.The network of about 400 nodes is created. The
transmission range is set to 250m. The nodes start
communicating with the nodes having packet in their buffers.
It is shown in fig3. The clustering is used to minimize energy
consumption. If cluster head drain its energy then member
nodes do join the new cluster by identifying the recent id of
the new cluster. The forms created are shown in fig4.
International Conference on Convergence of Technology – 2014 978-1-4799-3759-2/14/$31.00??2014 IEEE 4
Fig3. Network creation
Fig4. Clustering Model
TABLE I. SIMULATION PARAMETERS
No. of nodes 400
Initial energy 100J
Transmission range 250m
Sleep/active time 0.05ms
Duration of simulation 150s
Packet size 512 bytes
Data rate 250kbps
Routing protocol AODV
VI. CONCLUSION
In this paper, the existing MAC layer protocols have been
analyzed. The previous research work on preserving energy
efficiency is being presented. To reduce energy consumption,
the idea of about adapting the duty cycle is presented in this
paper. Working under variable traffic rates and multihop
network is a good idea that is the motivation got from the
previous research. From the proposed algorithm, it is expected
that energy consumption will be less as compared to the
studied methodologies. If there will be low traffic in the
network, no need to keep various configurations on each node.
For general routing, the clustering is the best approach used
for reducing energy consumption. With proposed
methodology, it is possible for the nodes in the cluster to
switch to a cluster that is more favorable and switch its radio
on only buffer contain packets. So, the same is used for
network having low traffic in the network. The proposed
algorithms will work on both cases i.e., for high as well as low
traffic. It is believe that the proposed scheme will significantly
better than the other schemes by keeping necessary nodes
alive; thus the network lifetime can be increased.
ACKNOWLEDGMENT
This is my final year ME research project in wireless
communication and computing under the guidance of V. A.
Gulhane. The work completed up to now is presented in this
paper. The work is still under implementation. The results will
be presented in the future. The research work in this paper was
supported by G. H. Raisoni College of engineering, Nagpur.
...(download the rest of the essay above)